Modelling of the Molecular Interactions in Solution Using Density Function Theory and the Conductor-like Screening Model Method

The "first principles" quantum mechanical methods are capable of correctly describing the electronic and geometrical structures of solute molecules. Traditionally most theoretical studies on molecules in gas phase have been carried out by ab initio methods or post-HF extensions in which electron correlation is included. Over the past decade the density function theory has emerged as a tangible and versatile computational method over the past decade. The advantage of DFT over ab initio method stems in its computational expedience and its reliability in most cases compared with the experimental results. Therefore, it is resonable to employ DFT as a practical tool to modelling molecular interactions in solution which demands large computational efforts.

Although DFT method can calculate considerably larger molecular systems such as transition metal complexes or large biological molecules, it is still difficult to include solvent molecules in the calculations explicitly beyond a limited number of solvent molecules.

The common strategy to avoid the difficulties mentioned above is try to directly introduce statistically averaged information on the solvent effects using a model called as dielectric continuum model which has suitable properties (dielectric constant, etc.). In this model the solute molecule is embedded in a dielectric continuum of permittivity. Thus the soluteforms a cavity within the dielectric and the surface charges are induced by the solvent molecules on the cavity surface. Because of the simplicity and its broad phsical meaning, the dielectric continuum model has gained many important progresses in recent years.

Recently, Klamt et. al proposed a new method within the dielectric continuum model called COSMO, based on the screening in conductors. COSMO method has two important features: one is that the surface-charge-solving process is non-iterative since it is included directly in the SCF procedure; the other feature is that it allows for the calculation of accurate gradients without cavity shape constriants which is significant for geometry optimization taking counts of solvent effects.

Because of the advantages of DFT on the calculation of solute molecule and the important features of COSMO method, it is very interesting to combine them together to treat the molecular interactions in solution.

There are some promising results of the application of COSMO method. All of their procedures used share at least a common defect, that is, the surface charges are obtained by using a so called matrix inversion method and this has the disadvantage of needing such a large memory requirment that the calculation for large molecules would be hard to perform from the practical point of view. We are planning to get around this problem by using the iterative method for the surface-charge-solving process. Besides, the program package (ADF) into which we are going to implement the COSMO method is unique since the core densities were kept frozen and the molecular densities were fitted by a set of auxiliary functions at sample points. This technique make implement COSMO method more flexsible and maybe more accurate since the surface charges induced by solvent molecules at cavity surface are distributed as point charges.

Back to Ziegler group Home Page